Scalable and Energy Efficient Computer Vision for Text Translation

Developments in cloud computing and smart phone technology have opened the door for many unique applications to be created. One of such applications is the ability to allow users to use computer vision with a camera on their phone to translate foreign signs into their native language. However, early adopters of this technology are far from optimal when it comes to features and robustness of their apps. Exploring options for optimizing allocation of resources and maximizing features of these apps can greatly improve the technology for users and distributors alike. In this paper we introduce a scalable and energy efficient computer vision protocol for the text translation to reduce power consumption, improving the data usage, and accuracy of translation. Our proposed idea is based on a camera driven process algorithm and an energy-efficient model to improve energy efficiency and provide the scalability support for foreign language translation. To validate the proposed idea, a Java based platform is developed. Furthermore, the performance of our application is test and compared with other applications of the same purpose. In the testing process, we randomly selected the set of words between languages. Our results demonstrate that our proposed energy efficient and scalable application showed much better performance than existing applications including google App.